Condensed from Paper No. 6 presented at CORROSION/96
Reproduced with permission from NACE International.

Current models often over-predict CC4 corrosion rates for wet gas and oil transport systems. One of the main reasons for this is the fact that the formation of corrosion product scales is not properly taken into account. An evaluation of in-house and literature experiments has been carried out to develop an equation for iron carbonate precipitation kinetics.

Modeling the kinetics of this precipitation process offers a number of applications in the area of corrosion prediction and control. The ultimate goal of the work is to establish the conditions under which stable, protective corrosion product layers form. A good indicator for the formation was found to be the so-called scaling tendency, which is defined as the ratio between the precipitation and corrosion flux. To form reliable scales, the bulk scaling tendency should be high enough for protective film formation and the local scaling tendency, in case of film damage, should be high enough for film repair. Both processes have been modeled and it was found that above approximately 80 ºC reliable scales are often easily formed whereas below this temperature, pH values of at least 6 are required.

The precipitation model also results in improved prediction of the solution pH which allows better corrosion rate descriptions even if the scales are not protective. In some cases the changes in pH may lead to reductions of 40% in the predicted corrosion rates. Other applications of iron carbonate precipitation modeling are also discussed.

Reliable prediction and control of corrosion is key to cost-effective and safe design of facilities for the gas and oil industry. As an example, current tools for the prediction of CO2 corrosion in pipelines" are still based on "worst case" assumptions, which may lead to unnecessary extra capital or operational expenditures to combat potential corrosion. These extra costs can stem from the use of expensive corrosion resistant steels, too much corrosion allowance (extra steel wall thickness), or application of a corrosion inhibitor. One area where there might be scope for design improvements concerns the description of the formation of corrosion product layers (scaling), which may limit the progress of internal pipeline corrosion. Currently the potential protective properties of scales cannot be taken into account in predictive models as adequate understanding of their stability and reliability, e.g. under fluid flow conditions, is lacking.

Test facilities

Experiments were carried out employing an autoclave, a low-flow loop with a fluid velocity of 0.1 m/s, and a high-flow loop with a velocity of 4 m/s.

Temperatures were between 40 ºC and 106 ºC, CC4 pressures between 0.55 and 5.6 bar. All experiments were performed oxygen free in a sodium chloride solution (1 wt%). The pH was continuously monitored in all experiments. In the 40ºC experiments the iron content was also frequently measured by photo-spectroscopy.

Twelve experiments were performed in the two flow loops at 40 ºC. In each flow loop, tests were performed for six different microstructures to examine the influence of flow and microstructure on carbonate precipitation: four normalized steels and two quenched and tempered. Chemical compositions and heat treatments are given in Table l.

Precipitation rate

The precipitation rate of FeCO3 was determined from the deviation from the linear increase of the iron counts with time during the corrosion experiments. This deviation starts after the saturation pH for FeCO3 precipitation is exceeded. The linear part of the curve is used to calculate the corrosion rate which was checked with weight loss measurements. The rate refers to the non-scaling part of the tests and is assumed to be constant over this period and during the early part of the subsequent precipitation period. Obviously, only the first part of the deviation from linearity can be used. Once the scale starts to become protective, the corrosion rate drops and it is no longer possible to determine the precipitation rate from iron counts.

ln theory the precipitation and corrosion rate can be derived from the increase of pH during the non-scaling period. In practice pH measurements are often too insensitive for this purpose, however. Experimental errors in the pH have been encountered and are most likely caused by scaling on the pH electrodes.

Table 2 lists the above-mentioned in-house experiments and relevant literature precipitation data cases. A distinction is made between data based on corrosion experiments with CO2and data based on chemistry experiments with salts (absence of metallic iron). All corrosion experiments are based on constant inventory tests with no net in- or out-flux of iron. Although additional experiments are available from literature, their results were not employed because of incomplete data (no pH and no iron counts or no corrosion rate), inaccurate data or inconsistencies in the data published.

The three experiments by Johnson and Tomson' are referred to as scanning experiments, in which the temperature was varied between approximately 35 ºC and 80 ºC. Since this makes comparison with other data difficult, it was decided to, artificially, split the experiments into "separate" experiments at 40, 50, 60, 70 and 80ºC and to derive precipitation rates for these temperatures.

In total 37 experimental cases were examined: 15 originating from the Johnson and Tomson data, 19 in-house experiments, one experiment by Dunlop et ql.' and two experiments by Dugstad'.

Reduction of corrosion due to FeCO₃ precipitation

The formation o. FeCO₃ layers will eventually lead to at least a global reduction of the corrosion rate. The exact corrosion reduction is difficult to predict in view of the many factors involved, like the type of steel, the flow velocity (both shear stress and mass transport effects), temperature, CO pressure, formation water composition, etc. It is clear that a full description of the influence of precipitation on corrosion rate is far too complicated. However, prediction of the corrosion rate reduction may be possible under specific conditions.

A further observation is that corrosion can only be reduced if the precipitation rate is of the order of the corrosion rate. If iron precipitation would be much more slowly than iron dissolution, the steel surface would be corroded away before a protective, dense layer could form.

The concept of scaling tendency is introduced here and is defined as the ratio between precipitation rate and corrosion rate (expressed in the same units). When the scaling tendency exceeds a critical value, a FeCO₃ film starts to grow and corrosion will reduce. This critical value is referred to as the critical scaling tendency. The small differences between the low and high flow experiments are probably not significant, also considering the fact that these experiments were performed in different flow loops with different surface/volume ratios, probably leading to slightly different critical scaling tendencies. The results show that the critical scaling tendency decreases when the carbon content in the steel increases. It is known' that carbon steels having a carbon content higher than approximately 0.15% can form a cementite/carbide network which remains on the surface after corrosion of the ferrite phase. This network can lead to a higher local supersaturation level, resulting in protective film formation at lower bulk supersaturation. Apart from the higher local supersaturation level, the type of corrosion reduction will also be influenced by the fact whether or not there remains a network on the steel surface. This effect will be discussed in the next sections.

For all experiments the critical scaling tendency ratio exceeds the value of 0.4. 1n the experiments with Armco iron and quenched and tempered steels at higher flow velocities no corrosion reduction was observed during the experiment (the maximum scaling tendency attained in these experiments was approximately 0.7).

Low carbon steels

The reduction of the corrosion process by scaling can be visualised in different ways. Localised FeCO₃sites develop first. Only at these sites corrosion is reduced. Assuming further coverage continues after earlier sites have reached a minimum scale thickness, the time for full coverage follows from the iron mass balance. During the second stage, the thickness of the layer increases but growth will continuously

slow down due to the corrosion reduction by the scale layer. As a minimum the model requires two parameters, which could be the layer thickness for full surface coverage and a layer thickness at which the corrosion rate is reduced to a certain fraction of the original rate.

Network-forming steels

For ferritic-pearlitic steels with more than 0.15% C, the corrosion reduction mechanism is different. Once the scaling tendency exceeds a critical value, the precipitation of iron carbonate in the pearlite network starts. The most simple description would be to assume that the corrosion rate is a linear function of the porosity. The porosity follows from the ferrite dissolution and iron carbonate precipitation in the network.

The corrosion rate in this study is assumed constant to limit the number of adjustable parameters. For the experiments discussed in this report a constant corrosion rate is an adequate approximation since most experiments were carried out at conditions which resulted in a reasonably pH-independent corrosion rate: high pH (scaling conditions) and low flow rates.

Open systems

In an open system (for example a pipeline) the change of the iron concentration with time is less relevant since the time required to develop a steady state situation is normally negligible compared to the lifetime of a pipeline. It is clear that the iron concentration and therefore the pH increase with distance as the iron concentration builds up. Due to the increased iron concentration and pH (e.g. carbonate concentration) the precipitation rate increases up to a point where the precipitation rate and corrosion rate balance (10 km), beyond this point a steady state situation is reached and the iron concentration and pH can only decrease when the corrosion rate drops due to protective film formation.

Closed Systems

In a closed system there is no net in- or out-flux of iron, leaving only the corrosion and precipitation terms in the iron mass balance. It is clear from the mass balance that after long exposure a kind of steady state situation (Fe_new=Fe_old) should be reached in which the amount of iron released by corrosion is equal to the amount of iron precipitated. In the case of the formation of a protective FeCO₃layer, the corrosion rate will tend to decrease to zero and the iron content will reach the saturation level. However, when no protective layer is formed the corrosion and precipitation rate will balance at a higher iron level.

Prediction of protective scale formation

The ultimate aim of our studies on precipitation kinetics is to enable the prediction of the formation of protective corrosion product layers and the resulting reduction of the corrosion rate. By combining models for corrosion rate and precipitation kinetics the appearance of FeCO₃layers can be predicted. To judge under what conditions full credit of protective iron carbonate layers can be taken, the following aspects should be clarified:

formation of protective corrosion product layers;

stability of these layers;

adherence to the steel surface of these layers;

repair of damaged scales.

The formation of protective scales can be described with the equations in this paper. The stability of the scales is relatively easy to predict; requirement for a protective stable scale is that the bulk conditions are at least saturated. No credit should be taken for scales which grow under local surface supersaturation conditions if the bulk remains undersaturated since this would mean that once fully protective films are formed and the local iron and carbonate gradients disappear (no corrosion), the scales would dissolve. The adherence of the scales is a function of various parameters like steel microstructure and flow rate. Some information has already been gathered but further work would be needed to fully cover this issue.

The last requirement is that a scale once damaged should repair reasonably quickly. This repair behaviour is no longer controlled by the bulk supersaturation since the bulk phase in a scaled system should be close to saturation. This means that repair should be possible at a low bulk supersaturation but probably a high local supersaturation at the damaged spot (inside a pit or mesa attack type of location). Modeling the local concentrations at the scratched surface to map the appearance of scale repair as a function of process parameters is discussed in the next section.

Improved pH prediction

The pH is an important parameter for corrosion prediction. Normally, the higher the pH, the lower the corrosion rate. The pH of a wet gas or oil/water transport system is often assumed to equal the saturation pH for FeCO₃precipitation. Since the kinetics of precipitation is very slow below say 60 ºC, the pH can reach much higher levels in practice (supersaturation) as discussed in this paper. At the beginning of a pipeline the iron build-up due to corrosion is relatively low and the pH will normally be lower than the saturation pH. By incorporating the FeCO₃precipitation kinetics, the pH along the pipeline can be calculated more accurately leading to a better corrosion prediction. The scale in these calculations is assumed non-protective.

Guidelines for inspection

More work will be required to establish the correct relationship between precipitation rate and corrosion reduction, but obviously, the higher the precipitation rate, the more likely the reduction of corrosion by corrosion product layer formation. A parameter which would be of direct use for operations already is the scaling tendency. By calculating the scaling tendency along a pipeline the critical parts (from a scaling point of view) of the system, having a low scaling tendency, can be predicted. These parts require special attention during inspection since the chance of insufficient protection by corrosion product scales is greater than for parts having a higher scaling tendency. The scaling tendency is low at the beginning of the line and around approximately 20 km. Along the line the temperature decreases leading to slower precipitation kinetics and the iron content builds up, leading to a higher driving force for scaling. These counter acting factors lead to the minimum in scaling tendency around 20 km. It weuld, therefore, be recommended to pay special attention to this part of the line during inspection and likewise to the very early parts of the line. By using the methods described above, installations which rely on the formation of protective scales can be monitored properly.

To improve on the prediction of CC4 corrosion rates, the kinetics of iron carbonate precipitation have been examined in detail, using both literature data and in-house experimental results. It was found that corrosion and precipitation experiments can be adequately described by using the proposed iron carbonate precipitation kinetics equation.

A model for evaluation of the repair characteristics of damaged scales as a function of operational and environmental conditions has been proposed. Calculations with this model show that damaged scales in condensed water systems will only repair at temperatures of at least approximately 80ºC. At lower temperatures scale repair in only possible if the pH is increased to values of at least 6.

Several applications of iron carbonate precipitation modelling like improved pH prediction and guidelines for inspection are also discussed.